We state the problem of inverse reinforcement learning in terms of preference elicitation, resulting in a principled (Bayesian) statistical formulation. This generalises previous work on Bayesian inverse reinforcement learning and allows us to obtain a posterior distribution on the agent's preferences, policy and optio…
On-device research index
arXiv research
A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.
169,181 papers · 148 categories
Trend · papers per month
48 results for “policy preferences”
This paper improves sample efficiency for off-policy evaluation with preference data.
problem Improving sample efficiency for off-policy evaluation with preference data.
method Using a deep neural network to learn the value function and leveraging manifold structure.
result Established a provably efficient guarantee for off-policy evaluation with RLHF.
New method optimizes policies without assuming known link functions between preferences and rewards.
problem Policy alignment with unknown and unrestricted link functions.
method Formulates an -divergence-constrained reward maximization problem, learning policies directly.
result Induces a semiparametric single-index binary choice model for policy alignment.
New framework estimates treatment effects based on preferences.
problem Estimating treatment effects with flexible outcomes.
method Preference-based Conditional Treatment Effect (CPTE) framework.
result CPTE provides interpretable targets and new identifiability conditions.
Paper analyzes finite-time guarantees for preference-based RL.
problem Understanding finite-time guarantees for preference-based RL.
method Combines dueling bandits and policy search to navigate state space.
result Identifies best policy up to accuracy ε with high probability.
A new approach to fine-tuning LLMs with human feedback.
problem Inability of current reward models to fully represent human preferences.
method Introducing NLHF, a new pipeline for LLM fine-tuning using pairwise human feedback.
result NLHF produces a sequence of policies converging to the regularized Nash equilibrium.
Study on identifying most preferred policy in bandits with vector-valued rewards.
problem Identifying the most preferred policy in bandits with vector-valued rewards.
method Derive a novel lower bound on sample complexity, design the Preference-based Track and Stop (PreTS) algorithm, and derive a new concentration inequality.
result The sample complexity of PreTS is asymptotically tight.
This paper analyzes MORL and proposes efficient algorithms to learn Pareto optimal policies.
problem Understanding and efficiently learning Pareto optimal policies in multi-objective reinforcement learning.
method Systematic analysis of optimization targets, reformulation of Tchebycheff scalarization, online UCB-based algorithm, preference-free framework.
result Identification of Tchebycheff scalarization as a favorable method and efficient algorithms for learning Pareto optimal policies.
Paper shows equivalence between two dividend preference models.
problem Understanding investor and firm preferences for dividends.
method Formulated Epstein-Zin preference, proved equivalence with Maenhout's model.
result Robust dividend policy is equivalent to a threshold strategy based on surplus process.
We use a simple agent based model of value investors in financial markets to test three credit regulation policies. The first is the unregulated case, which only imposes limits on maximum leverage. The second is Basle II and the third is a hypothetical alternative in which banks perfectly hedge all of their leverage-in…
Integrates multiple feedback channels into policy learning for reinforcement learning.
problem Combining multiple types of feedback into policy gradient algorithms.
method Lagrangian relaxation to satisfy constraints using gradient descent while maximizing rewards.
result Constraints are respected and can accelerate learning in reinforcement learning tasks.
SafeMIL learns safer policies by avoiding risky behavior from non-preferred trajectories.
problem Learning safe imitation policies from non-preferred trajectories in risky environments.
method SafeMIL uses Multiple Instance Learning to learn a cost function from non-preferred trajectories.
result SafeMIL learns a safer policy that avoids non-preferred behaviors without sacrificing reward performance.
f-GRPO and Beyond: Divergence-Based Reinforcement Learning Algorithms for General LLM Alignmentcs.LG
Extends reinforcement learning alignment to scalar rewards, improving math reasoning.
problem Designing reinforcement learning algorithms for general LLM alignment.
method Introduces f-GRPO and f-HAL, estimating f-divergences between reward-aligned and unaligned distributions.
result Improves math reasoning RLVR tasks and mitigates reward hacking.
Paper shows equivalence between two alignment methods and introduces a new algorithm.
problem Ensuring human alignment of large language models for useful, safe, and pleasant user experience.
method Introduces IPO-MD algorithm, showing equivalence between IPO and Nash-MD methods.
result Equivalence between IPO and Nash-MD methods proven when considering online version of IPO.
Paper uses neural word embeddings to analyze UN speeches for policy preferences and voting behavior.
problem Analyzing policy preferences and paradigm shifts in international politics.
method Applied neural word embeddings (Word2vec) to UN General Debate speeches.
result Found statistical relation between speech semantic content and voting behavior, contrary to hypothesis.
OSIL: Learning Offline Safe Imitation Policies with Safety Inferred from Non-preferred Trajectoriescs.LG
OSIL learns safe policies from unsafe demonstrations.
problem Offline safe imitation learning with implicit safety.
method Formulates CMDP, infers safety from non-preferred trajectories, learns cost model.
result OSIL learns safer policies without degrading reward performance.
We analyze the problem of learning a single user's preferences in an active learning setting, sequentially and adaptively querying the user over a finite time horizon. Learning is conducted via choice-based queries, where the user selects her preferred option among a small subset of offered alternatives. These queries …
PILAF optimizes reward models from human feedback for better policy alignment.
problem Creating accurate reward models from human feedback for policy optimization.
method Policy-Interpolated Learning for Aligned Feedback (PILAF) that explicitly aligns preference learning with maximizing underlying oracle reward.
result PILAF is optimal from both optimization and statistical perspectives, demonstrating strong performance in RLHF settings.
Zeroth-Order Policy Gradient for Reinforcement Learning from Human Feedback without Reward Inferencecs.LG
Paper develops methods to optimize policies directly from human feedback without reward inference.
problem Challenges in RLHF, including reward model overfitting and distribution shift.
method Develops two algorithms for RLHF without reward inference, using zeroth-order gradient approximators.
result Establishes polynomial convergence rates and outperforms existing methods in numerical experiments.
A new policy for contextual bandits adapts to reward vector shifts.
problem Learning under reward vector shifts with ordered rewards.
method Adaptive-discretization and optimistic elimination policy.
result Established upper bounds on preference-based regret.
Paper improves teaching by considering learner's preferences and constraints.
problem Teaching without considering learner's preferences and constraints.
method Design of learner-aware teaching algorithms that account for learner's preferences and constraints.
result Significant performance improvements over learner-agnostic teaching.
Provably Mitigating Overoptimization in RLHF: Your SFT Loss is Implicitly an Adversarial Regularizercs.LG
Mitigates overoptimization in RLHF by reformulating SFT loss as a preference optimization loss.
problem Overoptimization in RLHF where reward model misguides generative model.
method Proposes a theoretical algorithm that minimizes maximum likelihood estimation and reward penalty, reformulates as simple objective combining preference optimization and supervised learning losses.
result Improved performance of RPO compared to DPO baselines in aligning LLMs.
Study preference-based reinforcement learning in episodic kernel MDPs.
problem Learning from episodic human preferences in reinforcement learning.
method Developed preference-based value estimation and confidence sets for kernel-based MDPs.
result Proved high-probability regret bounds that converge to optimal policy value.
The paper sorts big data by revealed preferences, improving consumer and policy decisions.
problem Sorting diverse consumer preferences for big data objects like colleges.
method Endogenous weighting of revealed preferences, considering spillover effects.
result Consistent steady-state solution to counterbalance equilibrium.
Proposes a robust algorithm for aligning large language models with human preferences.
problem Misspecification in preference models, reference policies, and reward functions.
method Doubly robust preference optimization algorithm.
result Superior and more robust performance compared to state-of-the-art algorithms.
Introduces RPU to explain randomization preference in dynamic settings.
problem Explains preference for randomization in dynamic investment problems.
method Introduces recursive perturbed utility (RPU) to incorporate randomization preference.
result Proves RPU-optimal portfolio policy is Gaussian and can be expressed in closed form.
Paper tackles offline preference-based RL with human feedback.
problem Offline Preference-based Reinforcement Learning with preference feedback.
method Two-step approach: MLE for reward estimation and distributionally robust planning.
result First guarantee for learning any target policy with polynomial samples.
New method for efficient online exploration in RLHF reduces regret.
problem Efficiently collecting new preference data in RLHF to refine reward model and policy.
method Proposes a new exploration scheme that directs preference queries toward reducing uncertainty in reward differences most relevant to policy improvement.
result Establishes regret bounds of order for online RLHF, with polynomial scaling in all model parameters.
The paper optimizes pension policies with guarantees and sustainability constraints.
problem Designing optimal pension policies with guarantees and sustainability constraints.
method Dynamic utility model, stochastic domain, overlapping generations, time-consistent decision criterion.
result Optimal investment/pension policy computed for a general framework.
Bal-PM reduces preference labeling costs for LLMs.
problem Efficiently acquiring human feedback for preference modeling in large language models.
method Bayesian Active Learning with entropy maximization in feature space.
result Bal-PM reduces the number of required preference labels by 33% to 68%.
New method learns decisions from collective preferences without individual covariates.
problem Making decisions online without individual covariates.
method Collaborative filtering, matrix completion bandit, ε-greedy policy, online gradient descent, inverse propensity weighting.
result Method outperforms benchmarks and reveals new discoveries.
PBO framework optimizes latent preferences over multiple objectives.
problem Optimizing latent preferences with multiple conflicting objectives.
method Proposes DSTS, a multi-objective generalization of dueling Thompson sampling.
result DSTS outperforms benchmarks and provides asymptotic consistency.
Efficiently identifies good policies by choosing contexts for human feedback.
problem Efficiently acquiring human feedback for preference alignment in large language models.
method Formalizes active exploration as a dueling bandit problem and proposes an active exploration algorithm with a polynomial worst-case regret bound.
result Proposed method outperforms baselines with limited human preferences on various language models and datasets.
Derives a new objective to learn from human preferences without approximations.
problem Learning from human preferences through RLHF relies on approximations that can lead to pitfalls.
method Derives a new general objective PO that bypasses both approximations.
result Demonstrates the superiority of the new objective to Direct Preference Optimisation (DPO) empirically.
New methods improve LLM preference optimization by intelligently weighting multiple reference models.
problem Improving LLM preference optimization with multiple reference models.
method Introducing four new weighting strategies for multiple-reference preference optimization.
result All four new weighting strategies outperform current methods on preference accuracy.
New MAB model incentivizes user arm-pulling with self-reinforcing preferences.
problem Balancing exploration and exploitation in recommender systems with incentivized user preferences.
method Proposes a new MAB model with random arm selection and two policies: At-Least- Explore-Then-Commit and UCB-List.
result Achieves expected regret and expected payment over a time horizon .
Investor optimizes portfolio under dynamic risk preferences.
problem Optimizing investment under uncertain future risk attitudes.
method Developed a general equilibrium framework and solved for subgame-perfect equilibrium policies.
result Equilibrium policies include a novel hedging component to counteract anticipated risk aversion changes.
Active learning framework for optimizing human preferences in reinforcement learning.
problem Selecting most informative feedback for training models of human preferences.
method Proposes an active learning framework to collect preferential feedback online or offline.
result Errors in DPO logit estimates diminish with more feedback.
ZSPO optimizes RL from unknown link functions using human feedback.
problem Designing RLHF algorithms for unknown link functions.
method Zero-order policy optimization with human preference feedback.
result ZSPO converges to a stationary policy with a polynomial rate.
Investigates optimal pension policies in PAYG systems with forward utility and ageing population.
problem Optimal investment and pension policies in PAYG systems with sustainability and adequacy constraints.
method Non-zero volatility forward CRRA utilities, closed-form optimal policies, detailed numerical analysis.
result Characterization of optimal policies and detailed impact analysis under various scenarios.
A nonparametric pricing policy learns customer preferences from covariates without prior assumptions.
problem Lack of historical data limits personalized pricing for new businesses.
method Nonparametric pricing policy that clusters customers based on covariates and preferences.
result Regret of order for a finite horizon .
This paper improves decision support in multi-objective planning by better eliciting user preferences.
problem Determining optimal policies from user preference profiles in multi-objective decision making.
method Extending Gaussian process and pairwise comparison methods to multi-objective scenarios, proposing new ordered preference elicitation strategies.
result Proposed elicitation strategies outperform existing methods and users prefer ranking.
Improved SAC with AWMP for better control tasks.
problem Discontinuous and non-smooth optimal policies in reinforcement learning.
method Advantage Weighted Mixture Policy (AWMP) for SAC, learning state-specific weights.
result SAC with AWMP outperforms SAC in four control tasks.
Paper tackles optimal policy learning with observational data in multi-action scenarios.
problem Optimal policy learning in multi-action settings with observational data.
method Review of estimation approaches, analysis of risk preference, discussion of potential failures.
result Average regret of a policy with multi-valued treatment is contingent on the decision-maker's attitude towards risk.
Ad exchanges use CORP to set reserve prices against strategic buyers.
problem Setting optimal reserve prices in ad exchanges with strategic buyers.
method Proposes CORP policy to learn and set reserve prices robustly.
result Achieves sublinear regret in unknown noise distribution.
Paper addresses online alignment of large language models under uncertain preference feedback.
problem Online alignment of large language models with misspecified preference feedback.
method Formulates an oracle-robust objective as a worst-case optimization problem for log-linear policies, and develops projected stochastic composite updates.
result Shows that the robust objective admits an exact closed-form decomposition and achieves oracle complexity.
A new method for RLHF using proximal point Nash learning.
problem Capturing real human preferences in RLHF.
method Proximal point Nash learning, embedding self-play updates into a proximal point framework.
result High-probability last-iterate convergence for the combined method.
Unified approach to RLHF tackles uncertainty in reward function.
problem Uncertainty in reward function learned from human feedback.
method Value-incentivized preference optimization (VPO) that regularizes the reward function with value function.
result Theoretical and practical guarantees for both online and offline RLHF settings.